The most agnostic scheme for that I know of is RDF (https://en.wikipedia.org/wiki/Resource_Description_Framework)
IMO, its main advantage over committing to a specific SQL schema is that it can be used to produce PhD’s (yes, it would be cool if the store could figure out that its more efficient by a mile to store first and last names together, or that the typical queries run on the system would be zillion times faster if it had an index on user ID, but over half a century of research hasn’t gotten us there for SQL databases, so making that (somewhat) work if we give the system even less information is serious work)
Can you elaborate a bit on some example use cases?